Research Area:  Big Data
Proposed decades ago, k-means is still the most popular algorithm for clustering. Despite the drawbacks of k-means, its advantages make it most attractive. Several researches have been conducted to alleviate the problems of k-means.We suggest here some simple modifications to optimize kmeans for scalability without much sacrifice in the precision.Current shift in emphasis of data mining towards Big Data requires fast algorithms that can scale well. We propose an idea how time-tested techniques can be adapted to changing needs. The implementation results demonstrate the impact simple modifications can bring.
Keywords:  
Author(s) Name:  Akansha Agrawal and Shreya Sharma
Journal name:  International Journal of Computer Applications
Conferrence name:  
Publisher name:  IJCA
DOI:  10.1.1.695.4091
Volume Information:  Volume 120 – No.17, June 2015
Paper Link:   https://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.695.4091&rep=rep1&type=pdf